Flow Switch?

[Ares]

couldnt you get an electronic sensor to read if there is current draw on the pump line? if the pumps working there has to be flow. or atleast Id say its a safe assumption.

[dima y]

I belive the phrase goes like this:
"if you assume you make an ass out of you and me"

there are many things that could happen in which the pump would still be drawing current but the would be no flow. Some examples are impeller detaching from the shaft, a kinck in the tubing making a blockadge. Crack in the res that led to no water being pushed around.

If you dont want to have to babysit your rig then this is definatly something you want to have.

[Ares]

guess your right, might be a slightly cheaper safety check. impeller seems most likely, unlikely a tube would just randomly kink with no movement. and resevior, same thing,

[nexxo]

@dima_y: those are sweet flow sensors! As long as they don't restrict flow too much, you should be in business. Maximillium uses one in his rig; talk to him and see what he thinks of using one.

[Maximilium]

Wow, thats a nice price for 5 flow sensors

dima y, here is the diagram of how i conected my flow switch, I have nothing bad to say about this sensors.

With my computer the motherboard will shuttdown if the CPU fan signal is lost, not every motherboard out-there will do this, i have the Abit IC-7 Max 3. I dont know what MB you have, but I recomend you to test it first, while the compute runing, disconect the CPU fan cable, if the computer shuttdown in less than 3 seconds then you should be able to do this:

Also, I installed a CPU Delay Timer set for 1 second, to prevent the Motherboard to shutt down when start up, it will take about 1/2 sec. for the flow switch to activate after i turn on my computer, without it, when i start my computer the MB will think the CPU fan is dead and it will shutdown by default.




I hope it helps,
Max.

[dima y]

thank you very much for the input! I have been playing around with them for a while and figured out the N.O. and N.C. contacts. right now I am trying to figure out what the best way is (like you said some mobos dont support shutdown if the CPU fan is dead, i want it to be more or less universal).

It is definatly worth having the sensors there.

[Gooserider]

Just as a counter argument, do you really want to shut down the PC because the pump quit? Isn't the key issue that something is getting to hot? (like a CPU?) If so, why the concern with detecting pump problems? Wouldn't it be sufficient and more appropriate to look for CPU over heating? Figure out the maximum temp you should expect with a full load, then add 5-10*C and set up something to start a shutdown if the temp hits that point.

Per all the discussion I've seen, even total pump failure it takes several minutes for a CPU to heat a WC block from 'alarming' to 'panic' temps, plenty of time to run a shutdown or at least a panic power off. OTOH, a kink or other partial failure might still allow the system to run at elevated but still safe temps even if the flow isn't 'adequate'. A really well laid out system might even be able to get a safe stable temp through convection currents in the coolant w/ no pump at all.

Gooserider

[iggiebee]

Originally Posted by Gooserider

Just as a counter argument, do you really want to shut down the PC because the pump quit? Isn't the key issue that something is getting to hot? (like a CPU?) If so, why the concern with detecting pump problems? Wouldn't it be sufficient and more appropriate to look for CPU over heating? Figure out the maximum temp you should expect with a full load, then add 5-10*C and set up something to start a shutdown if the temp hits that point. Gooserider

That's exactly what I do on my 226W pelt + WC system, using MBM (Mother Board Monitor) and a little program called ShutDown Now. In my (particular) case, as soon as the CPU temperature hits +33C something really wrong has happened, and MBM is set to initiate a shutdown. Normal CPU temps for my "peltierized" computer should always be in the range 8C min to 28C max when overloaded.

I would be nice though to have some sort of flow data, although the ones I have seen so far are too restrictive.

[Maximilium]

[quote=Gooserider]Just as a counter argument, do you really want to shut down the PC because the pump quit? Isn't the key issue that something is getting to hot? (like a CPU?) If so, why the concern with detecting pump problems? Wouldn't it be sufficient and more appropriate to look for CPU over heating? /QUOTE]


I have the answer of that, NO.

Lets use a P4 for example, is supose to turn off the computer when it gets too hot, before it reach that point the prosesor is going to slow down ( with the posibility of corrupting any program runing).

Did you ever had problems turning off the computer on windows 98, xp, 2000 and the computer will hang because a program stop responding?

Thats why for some people (and myself) is very important to have a "fail safe", it will turn off the computer without depending on the software.

On my computer, I am running 2 mcp-600 pumps from an external 60w power supply, if that power suply goes bad, or a cable gets disconected, or for some reason the pumps stop working, my computer will shuttdown in 1 second.

[iggiebee]

Originally Posted by Maximilium

I have the answer of that, NO.

Lets use a P4 for example, is supose to turn off the computer when it gets too hot, before it reach that point the prosesor is going to slow down ( with the posibility of corrupting any program runing).

Did you ever had problems turning off the computer on windows 98, xp, 2000 and the computer will hang because a program stop responding?

Thats why for some people (and myself) is very important to have a "fail safe", it will turn off the computer without depending on the software.

In my case the "hardware" failsafe is pretty easy. Forgot to mention that I've set the BIOS controlled shutdown at a higher temperature and that takes care of things if the software controlled shutdown fails. Can't get any easier.

It all depends on your needs, and there are no best ways to do it. I personally use MBM + Shutdown Now because the computer I own is the one I use for my everyday work, and there is a *very important* need to save any open files during an emergency shutdown, and additionally avoid the high risk of disk corruption when suddenly shutting down a computer.

Only as a last resort, is the hardware controlled shutdown called in place.

[Maximilium]

[quote=iggiebee]In my case the "hardware" failsafe is pretty easy. Forgot to mention that I've set the BIOS controlled shutdown at a higher temperature and that takes care of things if the software controlled shutdown fails. Can't get any easier.QUOTE]

True, I also have set on my bios to shuttdown if it gets too hot.

This is the "thing", there is a lot of motherboards out that "are suppose" to shutt down if the fan fail, or if the cpu gets to hot.

Allmost all motherboards won't start if they don't register a cpu fan rpm signal or if the temperature of the CPU if out of range, but, after they boot up, you can actualy disconect the fan and the computer will stay on until it burns with the setings on at the bios (I won a case of beer on a bet I had with a friend, he didnt belive me).

Not all motherboards are the same, the only way to know is testing the motherboard. Not because I can turn on the "jump out of the cliff if CPU gets too hot " option on the bios means the computer actually is going to do that

[BladeRunner]

[quote=Maximilium]

Originally Posted by iggiebee

In my case the "hardware" failsafe is pretty easy. Forgot to mention that I've set the BIOS controlled shutdown at a
Allmost all motherboards won't start if they don't register a cpu fan rpm signal or if the temperature of the CPU if out of range, but, after they boot up, you can actualy disconect the fan and the computer will stay on until it burns with the setings on at the bios (I won a case of beer on a bet I had with a friend, he didnt belive me).

Not all motherboards are the same, the only way to know is testing the motherboard. Not because I can turn on the "jump out of the cliff if CPU gets too hot " option on the bios means the computer actually is going to do that

Yes I had exactly this with abit motherboards when I was making my shutoff system. I planned to use the "shutdown if cpu fan fail" setting in the bios but mine was even worse as it didn't even work when enabled on boot up with the fan disconnected. I even tried a stalled connected fan.

I also can't use the bios temp shut down because the lowest it allows is 50C and my CPU would have locked up before it got this hot as it normally runs in the 25C region. Like you I use mbm5 but with shutdown plus to force hibernation, so as hopefully not to lose any work in event of a shutoff. I still think it would be wise to have a hardware based "pull the power if the software hibernation doesn't work" backup, and I'll work on this in the future.

[Gooserider]

Maximilium:
GR: Just as a counter argument, do you really want to shut down the PC because the pump quit? Isn't the key issue that something is getting to hot? (like a CPU?) If so, why the concern with detecting pump problems? Wouldn't it be sufficient and more appropriate to look for CPU over heating?

I have the answer of that, NO.

Lets use a P4 for example, is supose to turn off the computer when it gets too hot, before it reach that point the prosesor is going to slow down ( with the posibility of corrupting any program runing).

Did you ever had problems turning off the computer on windows 98, xp, 2000 and the computer will hang because a program stop responding?

Actually, no, since I run Linux

However note that I did NOT rule out the 'jump off a cliff' power shutdown! I'm just suggesting that it be done with CPU temp rather than coolant flow. This is something that isn't that difficult, many fan controllers either have the ability to do a hard shutdown built in, or could be modded to do it very easily. (And no BR, you don't have to plug a fan into them )

Even running a stable and reliable OS like Linux, I still believe that a non-software / computer independent shutdown system is a vital part of good system design, I just think it should be operated by the key critical function, not something which might not matter as much.

Assuming the fail-safe fails safely, a CPU temp based hardware shutdown will trip on just about any failure, where there might be times a CPU would over heat w/o upsetting a flow sensor - for instance a block getting dislodged or a rad's airflow being blocked.

Gooserider

Gooserider

[dima y]

thats is why you do both!

[BladeRunner]

Originally Posted by dima y

thats is why you do both!

Exactly!

I think we've been down this road before, if you are not carefully it all gets overly complicated. I see cpu, (or any other items), heatup a secondary happening, whereas as the pump and it's flow the primary thing to monitor. Currently I have the flow monitored and the cpu temp similarly monitored with over temp shutoff by mbm. An emergency power down system is only as good as the last time you knew it worked. While I'd not deny something that pulls the plug is the best safety, it's not something I'd want to test on a regular basis for obvious reasons, and this is the only way to be sure it is still working for when you really need it to. This is why I favour a hibernation forced shutoff because it can be tested often without losing or possibly corrupting data / os. My experience using an upto date win xp pro install is it has never yet failed to force a hibernation within 30 secs, whatever I'm doing on the PC at the time including gaming, (I test it randomly a few times each week by kinking the inflow tube to stop flow).

I fully intend to have an instant power off system as a last resort probably temp triggered, but there are still limits as to what you can protect against. I still maintain the best prevention is largely shaped by the quality of the build and installation of the water-cooling system in the first place. There was a topic ages back detailing a psu hardware shut off circuit that monitored cpu temp to pull the power to the PSU in event of over temp. It's a good idea but was put forward in the topic that it could save the system from the block or heatsink falling off. I don't personally think it would as the core heat-up / compared to the psu actually being completely drained after power would still most likely fry the CPU, (assuming a p4 with thermal throttling wasn't used),

There will always be scenarios you cannot think of or prepare for, but I'm now almost 1/2 year past my 2 year major water-cooling disaster prediction point, shutoff system has never been required in anger... so personally I'm quite happy

[Butcher]

I just use the bios over-temp shutoff - it works (I've started my comp a fair few times without remmebering to turn the pump on - these days I just leave them pump on 24/7). The no-cpu fan is largely not an issue as you can disable it on most mobos that I've had.

BTW if your p4 slows down and corrupts programs you need to run better programs.

[nexxo]

Another scenario in pump failure is where stagnant water inside an overheating block starts to soften and possibly melt the tubing connecting to the block. The CPU may cope for a while, but some people have found that their mobo got a sudden bath. Best to have as many failsafes as possible. The circuit I'm putting in has the option of fitting a leak detector as well.

[Gooserider]

Leak detectors are good, and IMHO should be done as 'kill power NOW' when triggered. I would also make them fairly independent of the rest of the system. One approach I've seen that I liked involved using a common basement type leak detector (used to detect sump pump failures, etc.)

These work by having two electrodes with a short gap between them, if dry there is no electric flow, and all is well. If the electrodes get wet, current flows between them and triggers the alarm. In this case the user modded the electrodes by lining the bottom of his case with two peices of aluminum screen seperated by a paper towel. In case of flood, the towel soaked up the water, shorted the electrodes and set off the alarm. The alarm had a piezo screamer, and he just bridged a relay across the terminals for the power shutdown which didn't keep the screamer from working so the box essentially would turn itself off and scream for help at the same time.

Simple and elegant....

Gooserider

[Pug]

For reference, I remember posting in a similar thread a while back - http://forums.procooling.com/vbb/sho...5&postcount=53

Starbuck - I'm happy to forward Bladerunners switch if you do the deal (and need me to) otherwise, I have a few of the ones Nexxo linked kicking about somewhere....

[*klonk*]

Heres one i found and its not sold in large quantities - price is in Euro

http://www.elfa.se/elfa-bin/setpage..../en/130109.htm

Kinda expensive, and for the life of me i wouldent know how to connect it

[Jabo]

I have my system double TEC'ed (CPU&GPU). Recently while hooking up newe power supply I manged to power 226watter without pump running (comp was off at the time). 10 secs and I had explosive tubing failure due to boiled coolant and water everywhere.
That's why imo the only way to implement safety in TECed rig is to use flow switch to perform instant powerdown of the whole system in case of no-flow situation (no time for anything else before coolant boils and core melts down instantly due to no cooling at all (overvolted and oced CPU & GPU of course) giving your hardware an instant bath at the same time. If there is still power anywhere you may sit back and enjoy fireworks and then big hole in your credit card later on .
What I will do (have not done it yet but recent boiling experience prodded me in he right direction) is to hook high amp solid state relay to flow switch and main mains power supply lead. If there's no flow power is going to be cut to the whole thing. I will also incoropate a simple switch to let it back on-line after fixing what failed.
Just my two pennies here

[iggiebee]

Originally Posted by Jabo

That's why imo the only way to implement safety in TECed rig is to use flow switch to perform instant powerdown of the whole system in case of no-flow situation)

I respect your opinion but in my case all I have to say is "Thank God For Epox". My 8RDA+'s Bios has faithfully shut down the system prior to any disasters on two ocasionas before I started using mbm5 to shutdown the system via software. Now, in combination to with mbm5 + "Shutdown Now", the system has perfectly shut down by software on all occasions (at least 3 that I can remember)

Here in Southern Florida, power failures are very common during the summer season, so I have all systems I own on some from of UPS, except the one with the W/C+pelt, where the water pump itself is not under a UPS, since I have not been able to find a decent UPS that can handle the sudden inductive charge by the water pump during a power loss.